第 19333 題：估計量偏誤

原題與已知條件

共同有限樣本空間、逐theta sampling law與同一total estimator。操作=bias-family-readback；資料={"theta_values":[[1,2],[8,10]],"laws":[[[1,17],[2,17],[3,17],[11,17]],[[2,21],[3,21],[4,21],[12,21]]],"statistic":[0,1,9,10]}。請依 exact_ordered_tuple_declared_by_interface 欄序回答。

• G001 — 共同有限樣本空間、逐theta sampling law與同一total estimator。操作=bias-family-readback；資料={"theta_values":[[1,2],[8,10]],"laws":[[[1,17],[2,17],[3,17],[11,17]],[[2,21],[3,21],[4,21],[12,21]]],"statistic":[0,1,9,10]}。請依 exact_ordered_tuple_declared_by_interface 欄序回答。

定義與適用範圍

沿用原題明示的論域、假設與來源 MKO 定義。本題已完成第二對話數理與中英語義覆核，並由原始參數逐題獨立重算模型、答案與每個算術步驟；原題仍為 candidate。

• mko-estimator-bias

逐步推導

source operation=bias-family-readback;original snapshot and mathematics retained

explicit ordered source output fields=[parameter_bias_rows, boundary]

bias parameter0 full law mass := sum(1/17, 2/17, 3/17, 11/17)

bias parameter0 full law mass = 1

bias:original parameter SLOT=0;same outcome carrier Fin(4);ALL aligned law masses=[1/17, 2/17, 3/17, 11/17]

full sum=1;all masses nonnegative=是;qualified probability law;zero-mass outcomes retained

bias parameter1 full law mass := sum(2/21, 1/7, 4/21, 4/7)

bias parameter1 full law mass = 1

bias:original parameter SLOT=1;same outcome carrier Fin(4);ALL aligned law masses=[2/21, 1/7, 4/21, 4/7]

full sum=1;all masses nonnegative=是;qualified probability law;zero-mass outcomes retained

bias:parameter SLOT carrier Fin(2);common outcome carrier Fin(4);ONE total original statistic array=[0, 1, 9, 10]

T takes only the observation index,not the unknown parameter;SAME function at all parameter slots;its sampling law may still vary with the slot

original parameter SLOT order0..1;aligned exact theta_values=[1/2, 4/5]

declared target g(theta)=theta;indices are not targets;claims range over THIS supplied finite parameter model only

bias parameter0 original outcome0 weighted statistic := mul(1/17, 0)

bias parameter0 original outcome0 weighted statistic = 0

bias parameter0 original outcome1 weighted statistic := mul(2/17, 1)

bias parameter0 original outcome1 weighted statistic = 2/17

bias parameter0 original outcome2 weighted statistic := mul(3/17, 9)

bias parameter0 original outcome2 weighted statistic = 27/17

bias parameter0 original outcome3 weighted statistic := mul(11/17, 10)

bias parameter0 original outcome3 weighted statistic = 110/17

bias parameter0 same-law expectation := sum(0, 2/17, 27/17, 110/17)

bias parameter0 same-law expectation = 139/17

bias parameter0 expectation minus actual target := sub(139/17, 1/2)

bias parameter0 expectation minus actual target = 261/34

fixed original parameter SLOT=0;actual aligned theta VALUE=1/2;SAME complete law=[1/17, 2/17, 3/17, 11/17];same statistic=[0, 1, 9, 10];weighted terms=[0, 2/17, 27/17, 110/17]

E_at_slot[T]=139/17;Bias=E_at_slot[T]-ACTUAL theta value=261/34

bias parameter1 original outcome0 weighted statistic := mul(2/21, 0)

bias parameter1 original outcome0 weighted statistic = 0

bias parameter1 original outcome1 weighted statistic := mul(1/7, 1)

bias parameter1 original outcome1 weighted statistic = 1/7

bias parameter1 original outcome2 weighted statistic := mul(4/21, 9)

bias parameter1 original outcome2 weighted statistic = 12/7

bias parameter1 original outcome3 weighted statistic := mul(4/7, 10)

bias parameter1 original outcome3 weighted statistic = 40/7

bias parameter1 same-law expectation := sum(0, 1/7, 12/7, 40/7)

bias parameter1 same-law expectation = 53/7

bias parameter1 expectation minus actual target := sub(53/7, 4/5)

bias parameter1 expectation minus actual target = 237/35

fixed original parameter SLOT=1;actual aligned theta VALUE=4/5;SAME complete law=[2/21, 1/7, 4/21, 4/7];same statistic=[0, 1, 9, 10];weighted terms=[0, 1/7, 12/7, 40/7]

E_at_slot[T]=53/7;Bias=E_at_slot[T]-ACTUAL theta value=237/35

ALL fixed-parameter target/expectation/bias rows=[[0, 1/2, 139/17, 261/34], [1, 4/5, 53/7, 237/35]]

Bias is this complete indexed family;no averaging over parameters and no prior inferred

逐步反向核對

• S022 · premise_and_result_audit — 回到原始資料與本步條件核對，不把單向推論誤稱等價。
• S021 · premise_and_result_audit — 回到原始資料與本步條件核對，不把單向推論誤稱等價。
• S020 · premise_and_result_audit — 以精確有理數重算 bias parameter1 expectation minus actual target，不以小數近似替代。
• S019 · premise_and_result_audit — 以精確有理數重算 bias parameter1 same-law expectation，不以小數近似替代。
• S018 · premise_and_result_audit — 以精確有理數重算 bias parameter1 original outcome3 weighted statistic，不以小數近似替代。
• S017 · premise_and_result_audit — 以精確有理數重算 bias parameter1 original outcome2 weighted statistic，不以小數近似替代。
• S016 · premise_and_result_audit — 以精確有理數重算 bias parameter1 original outcome1 weighted statistic，不以小數近似替代。
• S015 · premise_and_result_audit — 以精確有理數重算 bias parameter1 original outcome0 weighted statistic，不以小數近似替代。
• S014 · premise_and_result_audit — 回到原始資料與本步條件核對，不把單向推論誤稱等價。
• S013 · premise_and_result_audit — 以精確有理數重算 bias parameter0 expectation minus actual target，不以小數近似替代。
• S012 · premise_and_result_audit — 以精確有理數重算 bias parameter0 same-law expectation，不以小數近似替代。
• S011 · premise_and_result_audit — 以精確有理數重算 bias parameter0 original outcome3 weighted statistic，不以小數近似替代。
• S010 · premise_and_result_audit — 以精確有理數重算 bias parameter0 original outcome2 weighted statistic，不以小數近似替代。
• S009 · premise_and_result_audit — 以精確有理數重算 bias parameter0 original outcome1 weighted statistic，不以小數近似替代。
• S008 · premise_and_result_audit — 以精確有理數重算 bias parameter0 original outcome0 weighted statistic，不以小數近似替代。
• S007 · premise_and_result_audit — 回到原始資料與本步條件核對，不把單向推論誤稱等價。
• S006 · premise_and_result_audit — 回到原始資料與本步條件核對，不把單向推論誤稱等價。
• S005 · premise_and_result_audit — 回到原始資料與本步條件核對，不把單向推論誤稱等價。
• S004 · premise_and_result_audit — 以精確有理數重算 bias parameter1 full law mass，不以小數近似替代。
• S003 · premise_and_result_audit — 回到原始資料與本步條件核對，不把單向推論誤稱等價。
• S002 · premise_and_result_audit — 以精確有理數重算 bias parameter0 full law mass，不以小數近似替代。
• S001 · premise_and_result_audit — 回到原始資料與本步條件核對，不把單向推論誤稱等價。

結論

([[0, 1/2, 139/17, 261/34], [1, 4/5, 53/7, 237/35]], bias_is_parameter_indexed_family)

依上列完整代值及條件核對得到原題答案；原題編號、格式與狀態保持不變。

本題覆核與來源綁定

覆核日期：2026-10-26

原題、答案、步驟表達式、數值 AST 與既有歷史保持原樣；原來的候選狀態屬於來源歷史，不覆寫為形式化證明。